Vimage compression based on multiscale ridgelet dictionary

Wavelets are well-known mathematical tools for representing one-dimensional (1-D) signals with a finite of discontinuities with a small number of coefficients. However, curve discontinuities are not fully captured by separable wavelets. In image compression application, high energy coefficients cluster around the edges and most of the bitrate is spent to code the contours. To capture the anisotropic geometrical structures, we firstly designed a multiscale ridgelet dictionary. It is anisotropic and directional and can capture geometrical patterns present in images. And then, we proposed an image coding scheme based on the matching pursuit (MP) expansion. To evaluate the proposed compression scheme, the compression performances of proposed method are compared against those of JPEG-2000. Experimental results show, at low and medium bit rates, the proposed method provides lower distortion and better visual quality.